Posted by Stephen Edie from ts3202.slip.colostate.edu (129.82.192.150)on May 03, 1998 at 06:11:15:
In Reply to: Redline entries. posted by Loser on May 02, 1998 at 05:02:30:
It depends on how you look at it. Quake
looks very nice on a Rendition card, but (on a
fast CPU) it runs faster...Sure... You forget
that almost all of those "BS" demos used
Zbuffering, blending, bilinear-filtering, and
simultaneous gourad-shading / texture-mapping.
With a highly optimized software renderer, those
effects combined running at a resolution of
640x480 generally run at around 5 seconds per
frame. ;)
If you disagree with me, you are welcome to
prove me wrong. When the source code of my entry
is released, write a good software renderer for
it. I'll appreciate it, especially if it runs
better and faster. :)
Remember that the point of the contest was
not necessarily to demonstrate effects that are
"only possible" on 3d hardware. I agree that
RRide was quite impressive. I've seen the same
effects implemented at full frame rate running on
slower Pentiums. On the other hand, RRide
presents these effects in a much cleaner fashion.
The source code probably looks like "bump-mapping"
code as opposed to "a mess of lookup tables and
realtime code generation routines". Indeed, with
enough programming effort, you could probably do
just about any effect, but every one else will
be laughing at you because they did it 2 years
ago with 3d hardware while you were still
optimizing yours beyond 1/10 fps...
One more consideration is interactivity.
Making a non-interactive demo is MUCH easier than
making an interactive demo. Adding interactivity
also cuts down on the number of 'tricks' one can
use to optimize the code. 3D hardware allows
us to develop today's games with yesterday's demo
effects.